Technology-Based Early Warning Systems for Bipolar Disorder: A Conceptual Framework
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
JMIR MENTAL HEALTH Depp et al
Viewpoint
Technology-Based Early Warning Systems for Bipolar Disorder:
A Conceptual Framework
Colin Depp1,2, PhD; John Torous3, MD; Wesley Thompson1,4, PhD
1
Stein Institute for Research on Aging, Department of Psychiatry, University of California, San Diego, La Jolla, CA, United States
2
VA San Diego Healthcare System, San Diego, CA, United States
3
Beth Israel Deaconess Medical Center, Department of Psychiatry, Harvard Medical School, Boston, MA, United States
4
Research Institute for Biological Psychiatry, Mental Health Centre Sct. Hans, Copenhagen, Denmark
Corresponding Author:
Colin Depp, PhD
VA San Diego Healthcare System
3350 La Jolla Village Drive (0603)
San Diego, CA,
United States
Phone: 1 8588224251
Fax: 1 8585345475
Email: cdepp@ucsd.edu
Abstract
Recognition and timely action around “warning signs” of illness exacerbation is central to the self-management of bipolar disorder.
Due to its heterogeneity and fluctuating course, passive and active mobile technologies have been increasingly evaluated as
adjunctive or standalone tools to predict and prevent risk of worsening of course in bipolar disorder. As predictive analytics
approaches to big data from mobile health (mHealth) applications and ancillary sensors advance, it is likely that early warning
systems will increasingly become available to patients. Such systems could reduce the amount of time spent experiencing symptoms
and diminish the immense disability experienced by people with bipolar disorder. However, in addition to the challenges in
validating such systems, we argue that early warning systems may not be without harms. Probabilistic warnings may be delivered
to individuals who may not be able to interpret the warning, have limited information about what behaviors to change, or are
unprepared to or cannot feasibly act due to time or logistic constraints. We propose five essential elements for early warning
systems and provide a conceptual framework for designing, incorporating stakeholder input, and validating early warning systems
for bipolar disorder with a focus on pragmatic considerations.
(JMIR Ment Health 2016;3(3):e42) doi: 10.2196/mental.5798
KEYWORDS
psychiatry; mHealth; prevention; technology; psychotherapy
them are a core element of many of the psychosocial
Introduction interventions for bipolar disorder [8].
The potential for technology to facilitate “early warning A promise of high frequency data collection agents, such as
systems” for bipolar disorder has been described for several mobile health (mHealth) technologies, is that potential warning
decades [1-3]. A number of reports have identified the existence signs and outcomes can be prospectively and concurrently
of near-term precursors of mood episodes [4-6]. These factors monitored. Moreover, predictions about the near future could
can be internal to the patient's “warning signs” (eg, disruptions be more accurate if based in part on accumulating knowledge
in sleep/wake cycles) or external factors or “triggers” (eg, life about a given patient, rather than upon more static risk factors
events) that are associated with increased risk for worsening that frequently fail to predict near-term trajectories for a given
course. It is apparent that warning signs are highly varied across individual. A variety of passive and active technologies have
individuals, although some data suggests they are consistent been piloted in bipolar disorder to this end [9-12], and while
within the same individuals over time (so called relapse the evidence base is limited and technologies seem better at
signatures) [7]. Due to the importance of early warning signs, monitoring depressive than manic symptoms [12], there is some
skills in monitoring and developing action plans to respond to
http://mental.jmir.org/2016/3/e42/ JMIR Ment Health 2016 | vol. 3 | iss. 3 | e42 | p. 1
(page number not for citation purposes)
XSL• FO
RenderXJMIR MENTAL HEALTH Depp et al
convergence of opinion that electronic self-monitoring will
become increasingly used in the management of this illness.
Proposed Architecture of Early Warning
Within a high frequency self-monitoring framework, an early
Systems
warning system might gather high frequency data on both We propose a conceptual framework for design elements of an
predictors (eg, sleep/wake cycle) and outcomes (eg, onset of early warning system. The following are the five elements of
mood episodes), identify when changes in sleep/wake cycles an early warning system (Table 1): (1) a platform or networked
that previously heralded a mood episode occurrence, and collection of platforms that enables frequent data collection (eg,
subsequently deliver an alert or intervention targeted to the early mobile phones, sensors), (2) inputs that produce intensive
warning sign in a timely fashion. Existing applications of early longitudinal data (eg, repeated self reports, collection of
warning systems are already part of daily life (eg, credit card behavioral data such as voice, sleep, activity) and outcomes that
fraud monitoring). Despite the potential of early warning could either be externally monitored events, states, or other
systems, there are a number of challenges to validation and also streams of intensive longitudinal data, (3) predictive statistical
potential harms. Problems may arise, both if early warning analytic methods that link inputs with outcomes, (4) decision
systems produce incorrect predictions and lead to unnecessary rules that determine the thresholds and actions according to the
distress or resource inefficiency, as well as if predictions are prediction based on a time horizon between the timing of the
accurate but patients or other stakeholders are unprepared or input data and the outcome that are predicted, and (5)
ill-equipped to act on warnings. This paper proposes a preventative feedback, which may vary by content, format,
framework for developing, validating, and implementing early delivery method, and intended audience.
warning systems with technologies that collect intensive
longitudinal data in bipolar disorder.
Table 1. Proposed components of an early warning system, selected techniques, and research gaps.
Component Selected resources Research gaps
Platform Mobile phone apps; text messaging; home-based telehealth; Best practices for long-term adherence and engagement; ef-
wearables fective integration of multiple platforms; user preferences
and methods for granular control of transmitted information
Inputs and outcomes Patient reports of mood and related risk factors; passive Predictive validity of near future and rare events; optimal
activity/location sensing; passive metadata sensing; serial data capture frequency and duration; interpretability of pas-
physiological sensors sive data for warning systems
Predictive analytics Linear and non-linear models for intensive longitudinal Integration of within-person and between-person data to in-
data; machine learning; within-sample and out of sample form predictions; integration of high dimensional variable
validation techniques frequency data; utility of non-linear, complex models in
practicable early warning systems; validation metric criteria
Decision rules Clinically important thresholds; empirically defined Developing interpretable decision rules based on multiple
thresholds based on classification models; recursive analy- inputs or interactions; updating decision rules based on accu-
ses to define earliest detectable change in risk at threshold mulating data within patients
Feedback Multiple communication platforms with which to alert Optimization of feedback messaging to enhance self-efficacy
stakeholders; elements of evidence-based behavioral change and health protective behavior; identification of potential
content developed for risk factor self-management in patient and other stakeholder’s experience of adverse impact
bipolar disorder of forewarnings; research methods for quantifying the impact
of individual feedback strategies; impact of feedback mes-
saging tailoring by mood state, patient preference, and/or
severity of risk
For the purpose of early warning systems, key requirements for
Platform Considerations platforms would be the availability of online statistical analysis
In considering the elements in Table 1, there is evidence to as data is accumulated, which typically and essentially if data
support the feasibility and acceptability of data capture platforms from other individual’s are used in prediction models, would
that use computer [13], mobile texting [14], wearable involve transmission of data from the device to a server for
technologies [15], and mobile phone apps in monitoring mood online analytics. Efficiency of transmission and interoperability
symptoms and other inputs in bipolar disorder [12]. Platforms of information sources gathered from the device or multiple
may increasingly combine passive sensor and device metadata devices remains a challenge that is especially pressing. Many
(eg, duration and timing of calls) with patient reported data. of the electronic self-monitoring tools for bipolar disorder
Advancements in software platforms for data collection are reviewed by Faurholt-Jepsen et al [12] were validated over
beyond the scope of this paper, but increasingly toolkits are periods of 3 months, which then necessitated that the predicted
available to generate apps that provide the interactive data outcome occur during that span, resulting in a low probability
collection framework on mobile devices such as with Apple of capturing multiple episode-level relapses for most patient
Research Kit and Android Research Stack. populations. Ideally such systems were to be developed to
capture more infrequent events over longer periods, and it is
http://mental.jmir.org/2016/3/e42/ JMIR Ment Health 2016 | vol. 3 | iss. 3 | e42 | p. 2
(page number not for citation purposes)
XSL• FO
RenderXJMIR MENTAL HEALTH Depp et al
unclear what platform-related factors contribute to longer-term understanding the clinical translation of such data is still an area
engagement (>3 months) in data collection. There is some of novel research and not fully understood.
suggestion that long-term engagement might be enhanced
Taken together, these early stage studies indicate that mobile
through interventions that promote continued use and perceived
devices could be feasibly used to monitor at least some of the
benefit, as well as the use of an individual's own device rather
potential inputs and symptom outcomes over time. Challenges
than study phones.
for inputs and outcomes in early warning systems are that
Since data is being transferred, which could include personally mHealth technologies create data that is complex and unique
identifying information (eg, phone numbers, location) as well from typical panel-type data, explained in terms of its high
as information about symptoms and other potentially sensitive volume, velocity, and variety [22]. The potential to collect large
information, mHealth platforms add a risk of loss of amounts of data must be balanced with adherence, missingness,
confidentiality. A recent review found that 75% of commercially interoperability, and validity concerns. Studies using mobile
available apps purported to assist with self-managing bipolar data collection have repeatedly shown that user adherence is
disorder did not include a privacy policy [16]. Given that not perfect and participants either completely stop using these
consumer buy-in for early warning systems would especially devices or provide data less frequently than planned. Indeed,
necessitate “trustworthiness” since the intent is to provide missing data has been attributed to a lower than anticipated
meaningful information about future personal risks, data security accuracy in a pilot study of the prediction of relapse in bipolar
procedures and risk from the device and/or sever to the cloud disorder using actigraphy [23]. Adherence issues may require
would need to be explained to users, including the potential robust statistical models for imputation [22] and may also be
consequences in loss of privacy if devices are lost [17]. When modeled as an additional input. Other issues with collecting
asked, patients express preferences for granular control over these new data steams the inter-stream and temporal correlation
elements of transmitted data and so apps for enabling such that will require partnerships with data science experts to fully
control, beyond just informing patients of the types of data being utilize and best understand the potential of real-time self report,
transmitted, may further facilitate patient engagement in early behavioral, and physiological data [24].
warning system design [18].
Moreover, some critically important outcomes and potential
Inputs and Outcomes new data sources would seem highly relevant to early warning
Potential inputs to early warning sign systems are far ranging systems but have received little research. In particular, suicidal
and include behaviors (eg, social behaviors, substance use, ideation and risk of self-harm has been effectively assessed over
changes in activity), sleep (eg, number of hours, quality), time in non-bipolar samples with attention to affective states
stressors, medication adherence, and affective states (eg, anxiety, prior to thoughts [25]. It is notable that patients may actually
irritability). These inputs measured with the device need to be be more open and willing to disclose suicidal thoughts to a
validated as convergent with gold standard clinical ratings. To digital device than in person to a clinician. Some people now
date, most validation studies in bipolar disorder using mobile also report suicidal thoughts and symptoms of their mental
devices have focused on the validity of mood ratings, and thus illness to online forums and their digital messages provide a
focus on the validity of measurement of the typical outcome of new source of data on suicide risk (eg, TalkLife), and a number
early warning systems rather than predictors. As reviewed of studies have linked risks of suicide from texts extracted from
recently, information technology strategies for self-monitoring social media and blog posts [26]. Data collection in the context
indicate positive indication of the concurrent validity of of an intervention that directly heralds suicidal behavior raises
aggregated momentary self-ratings of depressive mood states the need for a robust clinical response framework [27]. As
(although less consistently with manic symptoms) captured on detailed elsewhere, there are a number of ethical and privacy
mobile devices with clinician-rated data [12], with greater pitfalls in incorporating social media data into to clinical
intra-variability than paper-and-pencil mood charts [19]. Further decision making [28], and providing patients with granular
toward the path of an early warning system, integrated control over the inputs to early warning systems would seem
information from multiple sensors has accurately identified the to be especially applicable to social media.
presence of episodes among patients followed longitudinally Application of Prediction Models
[20]. It is unclear if patient reports of other inputs, such as
Prediction models link input data and subsequent outcomes. A
stressors, medication adherence, or social function are associated
variety of emerging methods are used to model intensive
with in-lab measures.
longitudinal data and predictive analytics applied to these
There are proof of concept data suggesting that passive sensor complex data that have varying rates and structures. Techniques
data obtained through actigraphy can individuate bipolar such as data mining, machine learning, and probabilistic
disorder from patients with depression and healthy controls modeling have been employed to make predictions about the
[21]. In addition, geographic or vocal tone correlates with future. While a discussion of these individual techniques is
concurrent clinician rated mood symptoms and metadata from beyond the scope of this paper (see [29]), effective early warning
interactions on the device are also moderately associated with systems would require automating analyses and responsive
symptoms [11]. While mobile phones and their multiple updating predictions based upon incoming data.
embedded sensors offer new streams of data, such as call logs
Irrespective of the statistical technique applied, validation of
to understand social patterns in bipolar disorder patients,
an early warning sign system would center on the accuracy and
preventative utility of prediction. There are several steps beyond
http://mental.jmir.org/2016/3/e42/ JMIR Ment Health 2016 | vol. 3 | iss. 3 | e42 | p. 3
(page number not for citation purposes)
XSL• FO
RenderXJMIR MENTAL HEALTH Depp et al
identifying a group-level association between T-1 predictor and decision rules can be developed empirically and a first step will
T-0 outcome. In terms of raising the clinical utility of predictions, require determining how sensitive and specific the input is in
the standard for gauging the usefulness of prediction would predicting subsequent outcomes, such as by use of a penalized
additionally include (1) how accurate the prediction is for a regression technique. Accuracy can be judged by a metric (eg,
given individual; (2) how interpretable the predictions are in area under the curve, AUC) generated by leave-one-out
the formation of decision rules; and (3) how much lead-time is validation, and can be compared to baseline estimates of risk
provided in which alter the course of the predictor, if alterable. (eg, most recent suicidal ideation rating), and/or relative to prior
windows of time that did not result in suicidal ideation, as in
Complex time series models have been applied to understanding the case-crossover method. If sensitive and specific to the
the dynamics of mood course in bipolar disorder, focusing on criteria listed above, sensitivity analyses to determine when
the potential for non-linear chaotic or latent approaches to prior to the outcome the earliest detectable increase in the
modeling state shifts in the context of intra-individual noise outcome occurs (eg, examining the accuracy of prediction by
[30-33]. Here we focus on potentially more accessible linear censoring predictors within the span of 1-3 weeks prior to the
models. In linear models, estimates generated from training data outcome), based on an assumption is that time-varying predictors
in ordinary least squares regression frequently result in increase in accuracy the closer in time to the outcome. The
overfitting, with validation samples highly likely to perform identification of the time of earliest detection would be
worse than the training set. Techniques such as penalized determined by comparison of accuracy when moving a stable
regression can reduce the likelihood of overfitting [34]. Given window of information back in time to when the prediction
that most samples are likely to be small at proof-of-concept accuracy falls below a clinical metric of accuracy (eg, AUC <
stages, validation with independent samples is typically 0.80) or a patient-preferred tolerance. To generate cut points,
impossible. Within sample alternatives to validation include regression tree methods can then be used to identify the levels
leave-one out validation [35]. of the input that would be associated with the best fitting model
Such models gauge the accuracy of predictions in samples with (ie, with the largest AUC). These cut points can then be used
and without predicted outcomes. However, this conflates to form decision rules and updated with machine learning
potential individual differences in the predictor-outcome algorithms in real-time, and made more precise to the individual
relationship. In repeated measures designs, case-crossover weighting information from the individual, relying increasingly
analyses can examine the association between predictors and on individual data over that from other individuals’ data.
outcomes within patients when outcomes occurred and earlier Multiple inputs and interactions among inputs may also create
or later times when outcomes were present [36]. This approach even more powerful models, yet with the tradeoff of greater
could help to identify the within-person sensitivity of predictors. complexity in implementation, and more importantly,
diminishing interpretability toward targeted feedback, described
As an example, Thompson et al [37] employed functional data next.
analysis to identify prospective time-lagged relationships
between changes in daily-assessed negative and positive affect Feedback and Clinical Application
and observed the emergence of suicidal ideation two to three The empirical understanding of best practices in the feedback
weeks later. Here, changes in affect can be seen as potential component of early warning systems is in its infancy. Since
early warning signs of increases in suicidal ideation. The authors early warning systems will involve communication around a
found that accuracy of the association between negative affect future probabilistic risk, there is substantial evidence to suggest
and suicidal ideation observed two to three weeks later was that people may variably interpret or misinterpret risk
quite accurate (88% sensitivity and 95% specificity), and more communication [38]. As such, the content and form of the
accurate than predictions that relied on baseline levels of suicidal feedback aspect of early warning systems is critical, and one in
ideation. In this way, indication of a time lagged relationship which perspectives of providers and patient stakeholders are
between inputs and outcomes suggests a potentially causal essential for understanding how best to communicate risk while
relationship and also provides a window of time between input mitigating potential adverse impacts, and maximizing
and outcome in which to deliver feedback regarding an tolerability, usefulness, and timeliness.
impending risk of suicidal ideation and possibly attempt to alter
the early warning sign. Feedback from early warning systems may be more effective
if it extends beyond simply notifying patients, clinicians, and
Creation of Decision Rules and Time Horizons stakeholders of an impending risk. There is substantial evidence
With validated data and prediction rules, a combination of that to change behavior in response to a future risk, the arousal
clinical acumen, patient preferences, and data driven models of fear or threat is insufficient and possibility counterproductive,
will be necessary to create decision rules that can guide clinical particularly among people with low self-efficacy to make
interventions. While some of these decision rules may be more changes [39]. For example, an early warning system describing
straightforward, such as a link between cessation of a medication a risk for a near-term manic episode might arouse fear about
and risk of switch into mania, others rules will be more complex. the consequences of mania, but without specific instructions
For example, considering insomnia as a symptom of bipolar about how to avert the manic episode, respondents may discredit
disorder, applying the right intervention at the right time and the message [40]. A related concern is that, if predictions derive
for the right patient will likely be a personal, dynamic, and from multiple sources of inputs including sensor data, it might
varied response with no clear-cut point or binary decision. Such be difficult to directly interpret; the drivers of risk may be
somewhat of a “black box” and may not translate to a greater
http://mental.jmir.org/2016/3/e42/ JMIR Ment Health 2016 | vol. 3 | iss. 3 | e42 | p. 4
(page number not for citation purposes)
XSL• FO
RenderXJMIR MENTAL HEALTH Depp et al
understanding of the illness course or actionable steps to mitigate like to tell themselves or adaptive behaviors they could employ
risk. Clinicians may face the same conundrums with acting on in depressed or manic mood states, which were then presented
the content of warning messages as do patients, additionally to them later through the device upon reporting such a state.
with the concern of balancing time allotted to care for patients Many other means providing user control via mHealth are
with current known risks versus future possible risks. possible, such as the frequency, timing, and communication
channel of feedback, as well as delineating whether and which
There are several potential approaches to enhancing the
other people are also recipients.
effectiveness of warnings derived from the broader health
psychology literature [41]. Messages may have greater impact Within early warning systems, there will be practical challenges
if they can winnow out actionable predictors that are meaningful to understanding the impact of various feedback strategies. As
to the individual, emphasizing one behavior change at a time. with all prevention research, it is challenging to determine
Once the target behavior is identified, messages may be more whether specific elements of early warnings systems have an
likely to lead to changes in behavior if they (1) focus on effect at the individual level, given that it is impossible to know
increasing self-efficacy to act and the benefits of acting rather if future threats would have occurred without such a warning.
than evoking fear about the risk of harm [42]; (2) identify As such, understanding the patient and message variables of
strategies that require low effort or facilitate reducing effort effective risk communication using prediction models may be
such as through implementation intentions [43]; and (3) provide best advanced in experimental studies with proxy measures of
affirmations of the individual prior to requiring the processing target behaviors rather than episodes, at least in initial
of threat, if delivering information about the threat is necessary development stages.
[44]. It is unknown how variation in mood state may impact the
receptiveness of different warnings, but it is technically feasible Discussion
for messages to be tailored to the mood state of the individual
to maximize persuasiveness. Summary
Designing the feedback element of early warning systems should Much work is required to make early warning systems accurate,
involve the incorporation of stakeholder perspectives, both useful, and safe for people with bipolar disorder and their care
patients and clinicians, which can occur at multiple levels. At teams. Nonetheless, there have been dramatic advances in
a broader level of intervention design, participatory design technology-based data capture, statistical prediction analysis,
methods have been evaluated in mental health and risk communication that together form the ingredients of a
technology-focused projects [45]. Although it is unclear if variety of early warning systems for bipolar disorder. Many
participatory approaches yield greater uptake and impact than such systems have been proposed and are in the proof-of-concept
researcher-centered programs, participatory methods typically stages of development, and soon will be available to consumers
include design cycles in which user input is sought at the outset and clinicians. These systems may make it possible for patients
and after multiple phases of development, from inception to to better understand and manage bipolar disorder, avoid or
deployment. For early warning systems, patient input about the forestall illness exacerbations, and minimize disruptions in
interpretability and timing of warnings along with content to social and productive roles associated with illness exacerbations.
accompany those warnings would be essential, as is clinician Conclusion
input about the liabilities associated with warnings. At the
We have described a basic framework for designing
narrower level of the early warning system design, there are a
interventions alongside patients and clinicians, and validating
number of methods for incorporating patient preferences and
and evaluating such systems. In particular, we caution that early
language into feedback. For example, patients may create their
warning systems must empower patients to make changes rather
own messages to accompany warnings for future display [9].
than to simply sound alarms. We hope that this paper stimulates
In our prior work, patients generated statements that they would
future development in this exciting area.
Conflicts of Interest
John Torous is editor-in-chief of JMIR Mental Health, but was not involved in the decision making process related to this paper.
The peer-review process was handled by a different editor.
References
1. Otto MW, Reilly-Harrington N, Sachs GS. Psychoeducational and cognitive-behavioral strategies in the management of
bipolar disorder. J Affect Disord 2003 Jan;73(1-2):171-181. [Medline: 12507750]
2. Depp CA, Moore DJ, Patterson TL, Lebowitz BD, Jeste DV. Psychosocial interventions and medication adherence in
bipolar disorder. Dialogues Clin Neurosci 2008;10(2):239-250 [FREE Full text] [Medline: 18689293]
3. Perry A, Tarrier N, Morriss R, McCarthy E, Limb K. Randomised controlled trial of efficacy of teaching patients with
bipolar disorder to identify early symptoms of relapse and obtain treatment. BMJ 1999 Jan 16;318(7177):149-153 [FREE
Full text] [Medline: 9888904]
4. Lam D, Wong G. Prodromes, coping strategies, insight and social functioning in bipolar affective disorders. Psychol Med
1997 Sep;27(5):1091-1100. [Medline: 9300514]
http://mental.jmir.org/2016/3/e42/ JMIR Ment Health 2016 | vol. 3 | iss. 3 | e42 | p. 5
(page number not for citation purposes)
XSL• FO
RenderXJMIR MENTAL HEALTH Depp et al
5. Lam D, Wong G, Sham P. Prodromes, coping strategies and course of illness in bipolar affective disorder--a naturalistic
study. Psychol Med 2001 Nov;31(8):1397-1402. [Medline: 11722154]
6. Perich T, Mitchell PB, Loo C, Hadzi-Pavlovic D, Roberts G, Frankland A, et al. Clinical and demographic features associated
with the detection of early warning signs in bipolar disorder. J Affect Disord 2013 Mar 5;145(3):336-340. [doi:
10.1016/j.jad.2012.08.014] [Medline: 22980404]
7. Miklowitz DJ. A review of evidence-based psychosocial interventions for bipolar disorder. J Clin Psychiatry 2006;67 Suppl
11:28-33 [FREE Full text] [Medline: 17029494]
8. Bond K, Anderson I. Psychoeducation for relapse prevention in bipolar disorder: a systematic review of efficacy in
randomized controlled trials. Bipolar Disord 2015 Jun;17(4):349-362. [doi: 10.1111/bdi.12287] [Medline: 25594775]
9. Depp C, Ceglowski J, Wang V, Yaghouti F, Mausbach B, Thompson W, et al. Augmenting psychoeducation with a mobile
intervention for bipolar disorder: a randomized controlled trial. J Affect Disord 2015 Mar 15;174:23-30 [FREE Full text]
[doi: 10.1016/j.jad.2014.10.053] [Medline: 25479050]
10. Faurholt-Jepsen M, Frost M, Ritz C, Christensen E, Jacoby A, Mikkelsen R, et al. Daily electronic self-monitoring in bipolar
disorder using smartphones - the MONARCA I trial: a randomized, placebo-controlled, single-blind, parallel group trial.
Psychol Med 2015 Oct;45(13):2691-2704. [doi: 10.1017/S0033291715000410] [Medline: 26220802]
11. Faurholt-Jepsen M, Frost M, Vinberg M, Christensen EM, Bardram JE, Kessing LV. Smartphone data as objective measures
of bipolar disorder symptoms. Psychiatry Res 2014 Jun 30;217(1-2):124-127. [doi: 10.1016/j.psychres.2014.03.009]
[Medline: 24679993]
12. Faurholt-Jepsen M, Munkholm K, Frost M, Bardram J, Kessing L. Electronic self-monitoring of mood using IT platforms
in adult patients with bipolar disorder: A systematic review of the validity and evidence. BMC Psychiatry 2016;16:A. [doi:
10.1186/s12888-016-0713-0]
13. Bauer M, Wilson T, Neuhaus K, Sasse J, Pfennig A, Lewitzka U, et al. Self-reporting software for bipolar disorder: validation
of ChronoRecord by patients with mania. Psychiatry Res 2008 Jun 30;159(3):359-366. [doi: 10.1016/j.psychres.2007.04.013]
[Medline: 18423616]
14. Bopp J, Miklowitz D, Goodwin G, Stevens W, Rendell J, Geddes J. The longitudinal course of bipolar disorder as revealed
through weekly text messaging: a feasibility study. Bipolar Disord 2010 May;12(3):327-334 [FREE Full text] [doi:
10.1111/j.1399-5618.2010.00807.x] [Medline: 20565440]
15. Valenza G, Nardelli M, Lanatà A, Gentili C, Bertschy G, Paradiso R, et al. Wearable monitoring for mood recognition in
bipolar disorder based on history-dependent long-term heart rate variability analysis. IEEE J Biomed Health Inform 2014
Sep;18(5):1625-1635. [doi: 10.1109/JBHI.2013.2290382] [Medline: 24240031]
16. Nicholas J, Larsen E, Proudfoot J, Christensen H. Mobile apps for bipolar disorder: a systematic review of features and
content quality. J Med Internet Res 2015;17(8):e198 [FREE Full text] [doi: 10.2196/jmir.4581] [Medline: 26283290]
17. Prasad S. Designing for scalability trustworthiness in mHealth systems. 2015 Presented at: Distributed Computing and
Internet Technology: 11th International Conference, ICDCIT 2015; 5-8 February 2015; Bhubaneswar, India p. 114-133.
18. Caine K, Hanania R. Patients want granular privacy control over health information in electronic medical records. J Am
Med Inform Assoc 2013 Jan 1;20(1):7-15 [FREE Full text] [doi: 10.1136/amiajnl-2012-001023] [Medline: 23184192]
19. Depp C, Kim D, de Dios LV, Wang V, Ceglowski J. A pilot study of mood ratings captured by mobile phone versus
paper-and-pencil mood charts in bipolar disorder. J Dual Diagn 2012 Jan 1;8(4):326-332 [FREE Full text] [doi:
10.1080/15504263.2012.723318] [Medline: 23646035]
20. Grünerbl A, Muaremi A, Osmani V, Bahle G, Ohler S, Tröster G, et al. Smartphone-based recognition of states and state
changes in bipolar disorder patients. IEEE J Biomed Health Inform 2015 Jan;19(1):140-148. [doi:
10.1109/JBHI.2014.2343154] [Medline: 25073181]
21. Krane-Gartiser K, Henriksen TE, Morken G, Vaaler A, Fasmer OB. Actigraphic assessment of motor activity in acutely
admitted inpatients with bipolar disorder. PLoS One 2014;9(2):e89574 [FREE Full text] [doi: 10.1371/journal.pone.0089574]
[Medline: 24586883]
22. Torous J, Staples P, Onnela J. Realizing the potential of mobile mental health: new methods for new data in psychiatry.
Curr Psychiatry Rep 2015 Aug;17(8):602 [FREE Full text] [doi: 10.1007/s11920-015-0602-0] [Medline: 26073363]
23. Novák D, Albert F, Španiel F. Analysis of actigraph parameters for relapse prediction in bipolar disorder: a feasibility study.
Conf Proc IEEE Eng Med Biol Soc 2014;2014:4972-4975. [doi: 10.1109/EMBC.2014.6944740] [Medline: 25571108]
24. Torous J, Baker JT. Why psychiatry needs data science and data science needs psychiatry: connecting with technology.
JAMA Psychiatry 2016 Jan;73(1):3-4. [doi: 10.1001/jamapsychiatry.2015.2622] [Medline: 26676879]
25. Muehlenkamp JJ, Engel SG, Wadeson A, Crosby RD, Wonderlich SA, Simonich H, et al. Emotional states preceding and
following acts of non-suicidal self-injury in bulimia nervosa patients. Behav Res Ther 2009 Jan;47(1):83-87 [FREE Full
text] [doi: 10.1016/j.brat.2008.10.011] [Medline: 19027892]
26. Won H, Myung W, Song G, Lee W, Kim J, Carroll BJ, et al. Predicting national suicide numbers with social media data.
PLoS One 2013;8(4):e61809 [FREE Full text] [doi: 10.1371/journal.pone.0061809] [Medline: 23630615]
27. Davidson CL, Anestis MD, Gutierrez PM. Ecological momentary assessment is a neglected methodology in suicidology.
Arch Suicide Res 2016 Jan 29:1-11. [doi: 10.1080/13811118.2015.1004482] [Medline: 26821811]
http://mental.jmir.org/2016/3/e42/ JMIR Ment Health 2016 | vol. 3 | iss. 3 | e42 | p. 6
(page number not for citation purposes)
XSL• FO
RenderXJMIR MENTAL HEALTH Depp et al
28. Wuyts P, Broome M, McGuire P. Assessing the mental state through a blog: psychiatry in the 21st century? The Psychiatrist
2011 Sep 30;35(10):361-363. [doi: 10.1192/pb.bp.110.030825]
29. Walls T, Schafer J. Models for Intensive Longitudinal Data. New York, NY: Oxford University Press; 2006.
30. Bonsall MB, Wallace-Hadrill SMA, Geddes JR, Goodwin GM, Holmes EA. Nonlinear time-series approaches in
characterizing mood stability and mood instability in bipolar disorder. Proc Biol Sci 2012 Mar 7;279(1730):916-924 [FREE
Full text] [doi: 10.1098/rspb.2011.1246] [Medline: 21849316]
31. Gottschalk A, Bauer MS, Whybrow PC. Evidence of chaotic mood variation in bipolar disorder. Arch Gen Psychiatry 1995
Nov;52(11):947-959. [Medline: 7487343]
32. Hadaeghi F, Hashemi Golpayegani MR, Murray G. Towards a complex system understanding of bipolar disorder: a map
based model of a complex winnerless competition. J Theor Biol 2015 Jul 7;376:74-81. [doi: 10.1016/j.jtbi.2015.02.020]
[Medline: 25728789]
33. Ortiz A, Bradler K, Garnham J, Slaney C, Alda M. Nonlinear dynamics of mood regulation in bipolar disorder. Bipolar
Disord 2015 Mar;17(2):139-149. [doi: 10.1111/bdi.12246] [Medline: 25118155]
34. Hastie T, Tibshirani R, Friedman J. The Elements of Statistical Learning. New York: Verlag; 2008.
35. Kohavi R. A study of cross-validation and bootstrap for accuracy estimation and model selection. IJCAI 1995:1137-1145.
36. Bagge C, Lee H, Schumacher J, Gratz K, Krull J, Holloman JG. Alcohol as an acute risk factor for recent suicide attempts:
a case-crossover analysis. J Stud Alcohol Drugs 2013 Jul;74(4):552-558. [Medline: 23739018]
37. Thompson W, Gershon A, O'Hara R, Bernert R, Depp C. The prediction of study-emergent suicidal ideation in bipolar
disorder: a pilot study using ecological momentary assessment data. Bipolar Disord 2014 Nov;16(7):669-677 [FREE Full
text] [doi: 10.1111/bdi.12218] [Medline: 24903771]
38. Gigerenzer G, Hertwig R, van den Broek E, Fasolo B, Katsikopoulos KV. “A 30% chance of rain tomorrow”: how does
the public understand probabilistic weather forecasts? Risk Anal 2005 Jun;25(3):623-629. [doi:
10.1111/j.1539-6924.2005.00608.x] [Medline: 16022695]
39. Heath RL. Best practices in crisis communication: evolution of practice through research. J Appl Commun Res 2006
Aug;34(3):245-248. [doi: 10.1080/00909880600771577]
40. Rimal R, Real K. Perceived risk and efficacy beliefs as motivators of change. Human Comm Res 2003 Jul;29(3):370-399.
[doi: 10.1111/j.1468-2958.2003.tb00844.x]
41. Ruiter R, Abraham C, Kok G. Scary warnings and rational precautions: a review of the psychology of fear appeals. Psychol
Health 2001 Nov;16(6):613-630. [doi: 10.1080/08870440108405863]
42. Basil M, Witte K. Health risk message design using the extended parallel process model. In: Cho H, editor. Health
Communication Message Design: Theory and Practice. Los Angeles, CA: Sage; 2012:41-58.
43. Gollwitzer PM. Implementation intentions: strong effects of simple plans. Am Psychol 1999;54(7):493-503. [doi:
10.1037/0003-066X.54.7.493]
44. Epton T, Harris P, Kane R, van Koningsbruggen GM, Sheeran P. The impact of self-affirmation on health-behavior change:
a meta-analysis. Health Psychol 2015 Mar;34(3):187-196. [doi: 10.1037/hea0000116] [Medline: 25133846]
45. Orlowski K, Lawn S, Venning A, Winsall M, Jones M, Wyld K, et al. Participatory research as one piece of the puzzle: a
systematic review of consumer involvement in design of technology-based youth mental health and well-being interventions.
JMIR Hum Factors 2015;2(2):e12 [FREE Full text] [doi: 10.2196/humanfactors.4361] [Medline: 27025279]
Abbreviations
AUC: area under the curve
mHealth: mobile health
Edited by G Eysenbach; submitted 28.03.16; peer-reviewed by N Berry, G Murray; comments to author 13.04.16; revised version
received 03.06.16; accepted 04.06.16; published 07.09.16
Please cite as:
Depp C, Torous J, Thompson W
Technology-Based Early Warning Systems for Bipolar Disorder: A Conceptual Framework
JMIR Ment Health 2016;3(3):e42
URL: http://mental.jmir.org/2016/3/e42/
doi: 10.2196/mental.5798
PMID: 27604265
©Colin Depp, John Torous, Wesley Thompson. Originally published in JMIR Mental Health (http://mental.jmir.org), 07.09.2016.
This is an open-access article distributed under the terms of the Creative Commons Attribution License
http://mental.jmir.org/2016/3/e42/ JMIR Ment Health 2016 | vol. 3 | iss. 3 | e42 | p. 7
(page number not for citation purposes)
XSL• FO
RenderXJMIR MENTAL HEALTH Depp et al
(http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work, first published in JMIR Mental Health, is properly cited. The complete bibliographic information, a
link to the original publication on http://mental.jmir.org/, as well as this copyright and license information must be included.
http://mental.jmir.org/2016/3/e42/ JMIR Ment Health 2016 | vol. 3 | iss. 3 | e42 | p. 8
(page number not for citation purposes)
XSL• FO
RenderXYou can also read
